2 When is a 2-Digit Number the Sum of the Squares of its Digits?

Transcription

1 When Does a Number Equal the Sum of the Squares or Cubes of its Digits? An Exposition and a Call for a More elegant Proof 1 Introduction We will look at theorems of the following form: by William Gasarch Theorem 1.1 Let L < U and k be in N. For every L x U there are XXX numbers that are the sum of the kth powers of their digits. The theorems we prove are easy and known. We will give several proofs of some of them with an eye towards the question: What is an elegant proof? For the last theorem I have a proof that is not elegant. If you know a more elegant one then please let me know. For each proof we note how many times we had to actually CHECK a number. This is a crude measure of elegance the fewer checks the more elegant the proof (compared to other proofs of the same theorem). We try not to game the system by having rather elaborate proofs that don t check that much. 2 When is a 2-Digit Number the Sum of the Squares of its Digits? We will prove the following theorem several different ways. Theorem 2.1 If 11 x 99 then x is not the sum of the squares of its digits. 2.1 Proof by Enumeration One can prove this theorem by listing all numbers between 11 and 99 and seeing that none of them work. One could write a program to produce this proof. We omit this proof. Number of Checks: : 89. PROS: You get the answer easily. If you wrote a program then you can adjust it to find the answers. to many more problems of this type (e.g., How many 3-digits numbers are the sum of the cubes of their digits) or variants of this problem (e.g., How many 2-digit numbers are within 1 of the sum of their digits? Which ones are they?) CONS: You get no insights. 2.2 Proof by Looking at each 10-number interval Case 1: x = 10a where 1 a 9 (so x ends in a 0). a 2 = 10a a = 10. 1

2 So this cannot happen. Note we did no checks. Case 2: 11 x 19. Since 11, 12, 13 > > > , none of 11,12,13 work. Since 15, 16, 17, 18, 19 < < < < < , none of 15,16,17,18,19 work. The only number left is 14 which does not work since Note we did one check. Case 3: Look at 21 x , 22, 23, 24 > , 27, 28, 29 < does not work since Note that we did one check. Case 4: 31 x , 32, 33, 34 > , 37, 38, 39 < Note that we did zero checks. Case 5: 41 x , 42, 43, 44 > , 47, 48, 49 < Note that we did one check. Case 6: 51 x , 52, 53, 54, 55 > , 57, 58, 59 < Note that we did zero checks. Case 7: 61 x , 62, 63, 64 > , 67, 68, 69 < Note that we did one check. Case 8: 71 x , 72, 73, 74 > , 77, 78, 78 < Note that we did one check. Case 9: 81 x , 82, 83, 84 > , 87, 88, 89 < Note that we did one check. Case 9: 91 x , 92 > , 96, 97, 98, 99 < , Note that we did two checks. Number of Checks: 8 PRO: Only ten cases. The techniques used may be helpful for other problems. CON: Only ten cases? If you do other problems with way the number of cases may get rather large. 2.3 Proof Using Mod 2, Mod 4, and Mod 10 Assume, by way of contradiction, that there exists 1 a 9 and 0 b 9 such that 10a + b = a 2 + b 2 henceforth the main equation. Take this mod 2 using x 2 x (mod 2) and 10 0 (mod 2) to obtain a 0 (mod 2). 2

3 If we take the main equation mod 4, using a 2 0 (mod 4), we get Hence b {0, 1, 4, 5, 8, 9}. We now take the main equation mod 10. a = b(b 1) 0 (mod 4). b a 2 + b 2 (mod 10). b(b 1) (mod 10). We now run through b {0, 1, 4, 5, 8, 9} and note for which ones a = b(b 1) (mod 10) exists and is a nonzero even number. Note that the squares mod 10 are 0, 1, 4, 5, 6, 9 so the only nonzero even ones are 4, 6. b b(b 1) (mod 10) a = b(b 1) (mod 10) comment NO GOOD: Need NO GOOD: Need NONE NO GOOD: No Square Root NO GOOD: Need NO GOOD: Need , 8 OH, Will need to check 9 8 NONE NO GOOD: No Square Root So the only candidates for (a, b) are (2, 4), (8, 4). One can check that these do not satisfy the main equation. Number of Checks: 2 PRO: The technique may extends to other problems. CAVEAT: Is this proof really more elegant then the intervals proof? 3 When is a d-digit Number the Sum of the Squares of its Digits? Theorem 3.1 If x 2 then x is not the sum of the squares of its digits. Case 1: 2 x 9. Since x < x 2 the theorem is true for 2 x 9. Case 2: 10 x 99. By Theorem 2.1 our theorem is true for these x. Case 3: 1 a 9, 0 b 9, 0 c 9. Assume x = 100a + 10b + c = a 2 + b 2 + c 2. The largest x can be is < 300. Hence a {1, 2}. The largest x can be is = 176 < 200. Hence a = 1 and b 7. The largest x can be is = 141. Hence and b 4. The largest x can be is = 98 <

4 Hence there is no such x. Case 4: x Let x = b m b m 1 b 0 where 1 b m 9, for 0 i m 1, 0 b i 9, and m 3. Since b m 1 we have x 10 m. Since x is the sum of the squares of its digits, x 81(m+1) Summing up 10 m+1 x 81(m + 1). We leave it to the reader to show that if m 3 then this is impossible. Number of Checks: 2 for Case 2, but 0 for the rest. So just 2. 4 When is a 2-digit Number the Sum of the Cubes of its Digits Theorem 4.1 If 11 x 99 then x is not the sum of the cubes of its digits. Assume, by way of contradiction, that there exists 1 a 9, 0 b 9 such that 10a + b = a 3 + b 3 henceforth the main equation. Hence 10a + 9 a 3 so a 3 10a 9 0. We leave it to the reader to show that this implies a 3. Take the main equation mod 2, using a 3 a (mod 2), b 3 b (mod 2), 10 0 (mod 2) to obtain Since 1 a 3 and a 0 we get a 0 (mod 2). (mod 2), we have a = 2. Plugging a = 2 into the main equation b(b 1)(b + 1) = 12 Alas, 12 cannot be written as the product of three consecutive numbers. Contradiction. Number of Checks: 0. 5 When is a 3-digit Number the Sum of the Cubes of its Digits Theorem 5.1 There is exactly one number 100 x 999 that is the sum of the cubes of its digits. That number is 153. Assume that there exists 1 a 9, 0 b 9, 0 c 9 such that 100a + 10b + c = a 3 + b 3 + c 3 henceforth the main equation. We will obtain conditions on a, b, c that force a = 1, b = 5, c = 3. 4

5 Claim 1: a b (mod 2). Proof of Claim 1 Take the main equation mod 2 to get a + b + c c (mod 2). Hence a b (mod 2). End of Proof of Claim 1 Claim 2: 1. b, c 7 2. If b 4 then c If c 7 then b Assume 3 a 8. If b 6 then c 5. Proof of Claim 2 Note that b 3 + c 3 10b c = 100a a 3. We will need the following table. a 100a a By the table if 1 a 9 then 99 b 3 + c 3 10b c 384. By the table if 3 a 8 then 273 b 3 + c 3 10b c ) If b 8 or c 8 then clearly b 3 + c 3 10b c 385. Hence b, c 7. 2) If b 4 then b 3 10b 24 hence 99 c 3 c Therefore c 3 c 75. Hence c 5. 3) If c 7 and b 5 then b 3 + c 3 10b c > 384. Hence if c 7 then b 4. 4) (3 a 8) If b 6 then b 3 10b 156 hence 273 c 3 c Therefore c 3 c 117. Hence c 5. End of Proof of Claim 2 Claim 3: 1. If a b (mod 4) then c / {2, 6}. 2. If a b (mod 4) and a is odd then c {2, 6}. Proof of Claim 3: 1) We take the main equation mod 4 using that a 3 b 3 (mod 4). 2b + c 2b 3 + c 3 (mod 4). 5

6 Since one of b, 1 b, 1 + b is even we have c 3 c 2b 2b 3 2b(1 b)(1 + b) (mod 4). c 3 c 0 (mod 4). This is not satisfied by any c {2, 6}. 1) We take the main equation mod 4 using that a 3 + b 3 0 (mod 4). 2b + c c 3 (mod 4). 2b c 3 c (mod 4). Since a is odd, b is odd. The set of c that satisfy this is {2, 6}. End of Proof of Claim 3 Claim 4: If a 0 (mod 2) then there is no solution to the main equation. Proof of Claim 4: If a 0 (mod 2) then, by Claim 1, b 0 (mod 2). The main equation mod 8, using that a 3 b (mod 8) and 10 2 (mod 8), is 2b c 3 c (mod 8). We now present a table of c 3 c (mod 8) as 0 c 7, and the possible values of b. c c 3 c (mod 8) b 0 0 0, , , , 4 For 0 c 6 we need, by Claim 2, b 6. Hence none of these values for c work. The only case left is c = 7. If c = 7 then 100a + 10b + 7 = a 3 + b a + 10b = a 3 + b Since 100a + 10b 336 we have a 3. Since a is even a 4. 6

7 For this item all mods are mod 10. If we take the equation mod 10 we get 100a + 10b = a 3 + b a 3 + b 3 4. a a 3 (mod 10) b 3 = 4 a 3 (mod 10) b The a = 8 case has c = 7 and b = 8 which cannot happen by Claim 3. candidates to check are (4, 0, 7), (6, 0, 7). None of them work. Hence the only Note: This claim took 2 checks. End of Proof of Claim 4 We can now enumerate all of the possibilities left. By Claim 4 we need only look at a 1 (mod 2). By Claims 1 and 2 we can assume b {1, 3, 5, 7} and c a = 1: By Claim 2 if b 4 then c 5. Since 6 3, we have c 5. Since we have that (a, b, c) (1, 5, 5). Hence the possible (a, b, c) are (1, 1, 5), (1, 3, 5), (1, 5, 0), (1, 5, 1), (1, 5, 2), (1, 5, 3), (1, 5, 4). By Claim 3 (1, 3, 5) and (1, 5, 2) is eliminated. Hence the only ones left to check are Only (1, 5, 3) works. Note: This took 5 checks. (1, 1, 5), (1, 5, 0), (1, 5, 1), (1, 5, 3), (1, 5, 4). 2. a = 3. By Claim 2 if b 5 then c 6. Since a (a, b, c) / {(3, 1, 6), (3, 3, 6)}. Since a (a, b, c) (3, 6, 6). Since a (a, b, c) / {(3, 5, 7), (3, 7, 4), (3, 7, 5), (3, 7, 6), (3, 7, 7)}. Hence the possible (a, b, c) are {(3, 1, 7), (3, 3, 7), (3, 5, 6), (3, 7, 0), (3, 7, 1), (3, 7, 2), (3, 7, 3)}. By Claim 3 (3, 1, 7) and (3, 7, 2) are eliminated. Hence the only ones left to check are 7

8 {(3, 3, 7), (3, 5, 6), (3, 7, 0), (3, 7, 1), (3, 7, 3)}. None of these work. This took 5 checks. 3. a = 5. By Claim 2 if b 5 then c 6. Since a (a, b, c) / {(5, 1, 6), (5, 3, 6), (5, 5, 6)}. Since a (a, b, c) / {(5, 7, 3), (5, 7, 4), (5, 7, 5), (5, 7, 7), (5, 7, 6), (5, 7, 7), (5, 3, 7), (5, 5, 7)}. The only cases left to check is (a, b, c) = {(5, 1, 7), (5, 7, 1)}. We can eliminate (5, 7, 1) by Claim 3. Hence the only case to check is (5, 1, 7). This case does not work. Note: This took 1 check. 4. a = 7. By Claim 2 if b 5 then c 6. Since a , (a, b, c) / {(7, 1, 6), (7, 1, 7)}. Since a , (a, b, c) / {(7, 3, 6), (7, 5, 6)}. Since a , (a, b, c) / {(7, 5, 7), (7, 7, 6), (7, 7, 7)}. The only cases left to check are (a, b, c) {(7, 3, 7), (7, 7, 7)}. By Claim 3 (7, 3, 7) and (7, 7, 7) are eliminated. Note: This took 0 checks. 5. a = 9. By Claim 2 if b 4 then c 5. Since a we have b 5 and c 6. Since a we have (a, b, c) / {(9, 5, 0), (9, 5, 1), (9, 5, 2), (9, 5, 3)}. The only cases left to check are (a, b, c) {(9, 1, 5), (9, 1, 6), (9, 3, 5), (9, 3, 6), (9, 5, 4), (9, 5, 5), (9, 5, 6)}. By Claim 3 (9, 1, 6), (9, 3, 5), (9, 5, 6) are eliminated. Hence the only cases left to check are (a, b, c) {(9, 1, 5), (9, 3, 6), (9, 5, 4), (9, 5, 5)}. None of these work. Note: this took 4 checks. Note: The total number of checks is 24. Question: Is there a proof of Theorem 5.1 with less checks? 8